开发者

Identifying swap sequences that produce the same result

开发者 https://www.devze.com 2023-04-11 18:44 出处:网络
I have an associative array where I store key-value pairs, on which I can perform swap operations: swap(a, b)// a <-> b

I have an associative array where I store key-value pairs, on which I can perform swap operations:

swap(a, b)    // a <-> b
  {
    temp = map.value(b);
    map.value(b) = map.value(a);
    map.value(a) = temp;
  }

Now, given a sequence of swaps, I would like to know if the next swap I perform causes the associative array to go into a state it was previously in:

e.g. the sequences:

1 <-> 2
2 <-> 1

and

1 <-> 2
2 <-> 3
3 <-> 1
2 <-> 3

both do nothing.

I want to be able to detect this by looking at the swap sequence itself. I'm guessing there is a mathematical formulation of problems of this kind, but I cannot seem to figure it out.

I observe that the first example is a loop and the second has a loop, so "detecting loops in a directed graph" might be part of the solution, but I'm not sure how this will fit into the algorithm I am looking for.

The algorithm should also work for unrelated swaps which are interleaved, the simplest example of this being:

1 <-> 2
100 <-> 200
2 <-> 1
200 开发者_如何学Go<-> 100


This is an application of permutations (and not really an application of graph theory).

Since these are key-value pairs, where presumably they could be sparse sets of numbers like your last example (1, 2, 100, 200), or even strings:

 Dancer <-> Vixen
 Prancer <-> Blitzen
 Comet <-> Donner
 Vixen <-> Rudolph
 Prancer <-> Dasher
 Comet <-> Cupid
 Vixen <-> Dasher
 Cupid <-> Donner
 Vixen <-> Blitzen
 Prancer <-> Dasher
 Dancer <-> Rudolph
 Prancer <-> Rudolph
 Comet <-> Cupid
 Prancer <-> Vixen

as well as numbers, what I would do is the following (there are at least two ways to do this):

  1. Scan the sequence to obtain a list L and a map ML of swap keys, such that L[ML[k]] = k. In the above example, L = [Dancer,Vixen,Prancer,Blitzen,Comet,Donner,Rudolph,Dasher,Cupid] and ML = {Dancer: 0, Vixen: 1, Prancer: 2, Blitzen: 3, Comet: 4, Donner: 5, Rudolph: 6, Dasher: 7, Cupid: 8}

Then either:

A. Procedural solution:

2A. Perform the swaps, in order, on the ML map itself. 3A. To check if the permutation leaves the map intact, verify that ML[L[i]] = i for each integer i between 0 and N-1 where N = the length of L.

B. Matrix solution:

Represent each swap as a permutation matrix, using the list L and map ML to convert from swap keys -> integer indexes. For example, Dancer <-> Vixen swaps elements 0 and 1, which is represented by the permutation matrix:

 010000000
 100000000
 001000000
 000100000
 000010000
 000001000
 000000100
 000000010
 000000001

) and multiply the permutation matrices. Then check if you have the identity matrix at the end.


The "A" solution is more efficient but the "B" solution with matrices is more mathematical in nature.

0

精彩评论

暂无评论...
验证码 换一张
取 消

关注公众号